Functional neurological disorder: new subtypes and shared mechanisms, 2022, Hallett, Dworetzky, Stone et al

Yes, the hypothesised bad prediction should give the opposite result on a comparison and update model.

In fact the whole predictive coding thing is pretty rubbish because it muddles up all sorts of different neural computational processes. The main predictive coding aspect is oculomotor. Our visual perception is entirely interpreted in the context of what they muscles are doing. Things being still is what we see if the retinal input is shifting in synch with our muscles moving the eyes. But even for visual perception there are other aspects that are not using predictive coding. For hearing it is all different at different stages. The blanket concept of predictive coding is just wrong. Herring a clap of thunder is not about your prediction of no clap of thunder being adjusted, you hear a clap of thunder!

But if predictive coding were involved, which it could be, its role in perceiving fatigue when all is well would be opposite.

It is quite intriguing how often one sees biomedical scientists making such basic mistakes in their theories. Which is why I query so many things around PoTS and salt intake etc. The theories don't actually predict what they think they predict. I have always been fascinated by these disconnects. The best one is the theory that in nephrotic syndrome the oedema is due to low oncotic pressure. But it cannot be because it occurs within hours of onset, long before plasma protein levels fall. It must be due to capillary or venular permeability changes.